- Email: [email protected]
Power switching with CdTe:Cl
304
Materials Science and Engineering, B I O( 1993 ! 304-308
Power switching with CdTe:C1 J. Lajzerowicz, L. Verger, F. M a t h y a n d M . C u z i n LETI (CEA, TechnologiesAvanc~es), CENG, BP 85X, 38041 Grenoble (France)
Abstract CdTe material with its high molecular weight (240) is today widely used for radiation detection. The b a n d - ~ p energy of CdTe is 1.45 eV. When doped with chlorine, which compensates the acceptor level introduced by cadmium vacancies, CdTe is intrinsic and gives a very high resistivity (more than 108f~ cm). The contacts were made by electroless metal deposition and further annealing; they were characterized with d.c. and pulsed voltages. For the first time, this paper presents power switching exp.eriments with CdTe material. We investigated different types of crystals and contact geometries with gap sizes varying from less than 1 mm to a few millimetres. The switches were activated by a YAG laser with 10 ns F W H M pulses ( 1.06/~m) or with 160 kV X-rays with 30 ns EWHM pulses. The time constant of recovery was found to be more than 10 ns. In some cases, for high voltages, corresponding to fields higher than a few kilovolts per centimetre, large recovery times of more than 100 ns were measured. This apparently Iolager carrier lifetime, combined with the high resistivity, make CdTe an alternative material to silicon and GaAs for some swishing applications.
1. Introduction--main properties of CdTe Semiconductor photoconductive switches have become increasingly useful for pulsed-power applications where high voltages must be switched on a short time-scale. Photoconductive switches can switch tens of kilovolts with picosecond rise times [1]. Most of the research has been concentrated initially on silicon and more recently on GaAs as the switch materials. Different materials, such as diamond and ZnSe, have also been evaluated for this application. To enlarge the group of potential semiconductor for the application, we decided to evaluate CdTe. CdTe has been recognized for several years as being the most promising detector material for a room temperature operating portable X-ray and gamma-ray detection system. It is also used as a substrate for epitaxial growth of H g I _xCdxTe to produce IR detectors. Semi-insulating CdTe is usually obtained by impurity compensation with chlorine or indium. Different techniques, such as the Bridgman technique or
travelling heater method (THM), are used to grow the crystals. Of course, the availability and purity of the crystals grown today cannot be c o m p ~ e d with those of silicon and GaAs. Table 1 gives the main characteristics of CdTe compared with silicon, GaAs and diamond. Different properties of CdTe must be pointed out as follows. CdTe has a direct b0JlO gap (Gunn effect is reported [2]). The mobility curve of CdT~ presents a maximum in velocity for a feld of the order of !5 kV ~m -~ [3]. Compared with GaAs and !nP, and following some of the given interpretations of the lock--on phenomenon [4], it should present the lock-on p r o p ~ y above 15 k V c m - L (The lock-on property is a non-linear effect: when the field applied is sufficiently high, instead of following the excitation pulse shape, the switch turns on and stays on until the energy is completely discharged from the generator.) CdTe recovery time is in the range right between the recovery times of silicon and GaAs.
TABLE 1. Comparison of main parameters for CdTe and the three main semiconductor materilals
Band gap (eV) Resistivity (~ cm) Mobility (cm 2 V- ~s- 1) Breakdown field (kV cm- 1) Recovery time (/~s)
0921-5107/93/$6.00
Si
AsGa
Diamond
CdTe
1.12 (indirect) 104 1000 105 10-100
1.43 (direct) 10 s 5000 105 0.001-0.01
5.47 (indirect) 10 l 1000 107 < 0.001
1.47 (direct) 109 1000 105 0.01-1
© 1993 - Elsevier Sequoia. All rights reserved
J. Lajzerowicz et al.
/
It seems very interesting to evaluate this material for power switching applications. We will briefly report how we developed ohmic contacts for our CdTe material. We will present the hold-off characteristics for impulsed voltages and, finally, we will show some results obtained when switching with laser or X-ray pulses.
2. Device preparation and characterization The monocrystalline and large volume CdTe crystals have been grown by the vertical Bridgman method by normal freezing of a tellurium-rich solution at a temperature lower than the melting point of stoichiometric CdTe [5]. Semi-insulating CdTe is obtained by impurity compensation: in the accepted scheme, a shallow donor, i.e. chlorine, is introduced into the material to compensate accepter native defects which are generally thought to be cadmium vacancies. By the selfcompensation phenomena, the balance between the chlorine donor and accepter concentration leads to Fermi level pinning within the gap that makes the material semi-insulating and thus greatly reduces photoconductor dark current [6]. Hall effect measurements exhibit a p-type conduction for the high resistivity chlorine-doped crystals. At room temperature, crystals have a resistivity p = 109 if2 cm, a hole concentration of 108 cm- 3 and a Hall mobility /~H = 50 cm 2 V-l s-1. (The Hall mobility is not the electron mobility and is much closer to the hole mobility.) We used two types of samples: one was 10 mm × 15 mm × 1 mm with the two contacts on the same large side separated by a gap of 3 mm and the other 10 mm × 10 mm x t mm (t is from 600/~m to 2 mm) with circular contacts on the opposite large sides. As with most highly intrinsic semiconductors, it is difficult to produce ohmic contacts on CdTe:C1. Our philosophy of experimentation was to carry out reliable, stable and reproducible ohmic contacts in order to obtain the lowest current for the highest electrical field applied when doing I ( V ) characteristics: 1 kV. This was obtained with electroless deposited platinum on mechanically lapped surfaces. (The surfaces are mechanically lapped to reduce the leakage current.) The solution was prepared by dissolving H2PtCI 6 in deionized water. The approximate thickness is 100 nm. Figure 1 shows the I(V) characteristics obtained for a 10 mm x 10 mm x 1 mm sample (contacts on opposite sides) before subsequent annealing. For applied voltages in the range from - 1 kV to 1 kV, the characteristic is linear and symmetric, i.e. ohmic behaviour. The sample impedance is found to be between 1 and
Power switching with CdTe.'CI
305
2 Gg2, which corresponds to a resistivity of a few gegaohm centimetres. Unfortunately, under d.c. voltages corresponding to 15 kV cm- ~most of our samples fractured. This corresponds to a rapid increase in the dark current at constant voltage. This is probably due to an increasing space charge limited conduction because of trapping on deep levels. Two different experiments were performed to test the samples under pulsed voltages. (1) Polarization by a 50 ns, 0-10 kV pulse from a line discharge (sample placed in a SF6 pressure chamber). For the 10 mm x 10 mm × 1 mm sample (contacts on opposite faces), appreciable conduction appears at 3.5 kV while at 10 kV the sample presents a 100 Q resistance. This corresponds to 20 mJ dissipated in the sample during 50 ns. For the 10 m m × 15 m m x 1 mm sample (contacts on the same side), conduction appears at 5.5 kV and the apparent resistance of the sample at 10 kV is 150 f2. After this test the I(V) characteristic of the 10 mm × 10 mm × 1 mm sample Was degraded. (2) Polarization by the 0.5/~F capacity discharge in 25 Q with 0-3 kV. The samples tested are those with contacts on opposite faces. With two contacts of the same dimensions, breakdown appears at about 22 kV
|
1,5
Y
I 0,5
--" C,)
13
-I -I ,S I~I -1000
,
.1500
11
0
I 5400
Before annealing After annealing
1000
VOLTAGE (V)
Fig. 1.1(V) characteristicsof a 10 mm × 10 mm x 1 mm sample.
Fig. 2. Sample top viewwith two breakdown paths.
306
J. Lajzerowicz et al.
/
l'ower switching with ('dl'e : (7
cm-J (1.5 kV for the 680/~m gap and 1.9 kV for the 850 # m gap). The paths are situated right along the contacts as it is shown on Fig. 2. They are holes in which the material melted. The sample shown presents two paths because, after several breakdowns in one path, the metal proceeds to sublimate and there is no more metallic continuity to the first path. The apparent breakdown resistance observed at the first breakdown at 1.5 kV (gap 680 ¢tm) is about 80 f2, which corresponds to 14 J dissipated in 10 kts, i.e. enough to melt CdTe. The paths go exactly from one contact frontier to the other contact frontier on the opposite side (see Fig. 3). The simplest solution to overcome such behaviour was to use a small contact on one face and a big contact on the opposite face (typically diameters of 2 and 7 mm on the 1 mm x 1 mm faces). Indeed, in this case, the breakdown field doubled to reach 40 kV cm- ~ (2.5 kV for the 620/~m gap and 3 kV for the 750/~m gap). The paths in this case go from the small contact frontier straight to the middle of the other contact (see Fig. 3).
dose at 1 m is 0.7 mrad. Figure 6 shows a diagram of the experimental set-up for X-ray switching under 0-3 kV pulsed voltages. The voltage pulse triggers the 300 kV high voltage X-ray tube supply; the delay is less than 1 ,us with a jitter of less than 200 ns. The time constant of the V0 pulsed voltage is 12.5/~s; therefore, during the X-ray pulse, the voltage across the switch is about 95% of V0. We only performed tests with samples with contacts on opposite faces. Switched signals give recombination times of the order of 20 ns for irradiation perpendicular to the contact and, in some conditions of irradiation from the
v
> t 206v~ I1
~
-'~-~-.
3. Results with laser or X-ray activation 200 ns/div.
We used a YAG YG 480 laser from Quantel. The energy delivered at 1.06/~m during the 12 ns pulse is 320 J. We used only d.c. polarization for laser switching tests. Different behaviours were observed in the recombination time which seems to depend on the geometry of the switch and on the field. Figure 4 gives an example of recorded switch signals at different fields. For 2.5 kVcm -~ and above, the apparent recombination times reach nearly 1/~s. Figure 5 gives the variation of the switch resistance vs. the beam attenuation at different voltages. The limit resistance seems to be lower at low fields. When the attenuation is high enough, the variation of the resistance is quasi-linear with the attenuation, with the same linearity whatever the field is. Assuming a mobility of 1000 cm 2 g -l s -1, we can calculate the equivalent energy deposited in the switch for each attenuation. We can then determine roughly the absorption of the beam to find between 0.5% and 1%. The X-ray generator is an HP 43733A flash generator. The maximum X-ray energy is 300 keV, the mean energy is 140 keV, the pulse width is 30 ns and the Same size e l e c t r ~
15"30V~
t 400 ns/div. Fig. 4. Signals from laser switching at the same attenuation but at different voltages (contacts on the same face (2 rmm gap) and laser beam on the opposite face) (recorded on Tektronix TDS 500).
10000
~-~1000 i
1 O0
#
10
Different size
1
Brea~uow. m ;~ ~ v/~=.
Breakdown at 40 K v/~m
Fig. 3. Breakdown paths and values for different contact geometries.
f, 0,05 kV/cm 0,5 kV/cm 2,5 kV/cm ........
10
t00 Beam attonuatlon
8
5 ky/cm
1000
Fig. 5. Switch resistance vs. laser beam attenuation (contacts on the same face and laser beam on the opposite face).
J. Lajzerowicz et al.
/
307
Power switching with CdTe : CI
HP 300 kV X-ray flash generator
I
10 I ~
I.tt.n..tor I ..h.°,o., I M I
"'"
I I
switch
o-3kv
/
I
.w,°,
.ttenuator
T ""T
,'°
L,';.n.°°°.,., I : TektronixI'
LI,o.o I
OSAOO'
Fig. 6. Experimental setting for switching tests with X-rays.
sides of the sample, the recombination times reach 400 ns. O n Fig. 7 is plotted the evolution of the switch resistance with the inverse of the irradiation dose at different fields. T h e irradiation dose is proportional to l i d 2 with d the distance to the source. Computation gives a carrier generation rate of 7 x 1011 c m - 3 s 1 at 1 m and, therefore, a dose of 1 corresponds to 3 × 1013 cm-3 s-1. F r o m this value and from the slope of the linear variation of the resistance, we can estimate the mobility, and we find 1150 cm 2 V-1 s-1 which is in agreement with our theory. We were able to switch at a field of 38 kV c m - l without observing any lock-on phenomenon. -
10 s ~ 104 8 ~ 103 i ~" 102 10 1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
,o
=
.
.
.
.
loo
looo
1/(Normalized dose Irradiation)
(a)
......
L
................ / f
8i-
_I
4. C o n c l u s i o n s
T h e aim of this work was to evaluate CdTe for power switching applications. We did not obtain answers to all the questions but what we are able to say is that CdTe can withstand 40 kV c m - l and does not present any lock-on at 38 kV cm-1. This breakdown voltage must be associated with the contact technology we used and should be improved when optimizing the technology and adapting the geometries. We were surprised not to see the lock-on and this result must be taken into account by scientists who want to interpret the lock-on phenomenon.
": .... A~plie~l ~iild "=""g kV]cm ..... Applied Field = 1,2 kV/~,
102
(b)
~ A p p l i e d field = 16 kV/cm Applied field = 35 kV/cm ........ = ........
10 100 1/(Normalizeddose Irradiation)
Fig. 7. Switch resistance vs. 1/(X-ray dose irradiation). (a) D.c. voltages on i mm x 1 mm x 0.8 mm sample. (b) Impulse voltages on 1 mmx 1 mm x 0.6 mm sample (contacts on opposite faces for both samples; contact sizes are different for the second sample).
M. C. Gentet, O. Maurice and J. Rustique for technical assistance.
References Acknowledgments
T h e authors want to thank B. Schaub and G. G a u d e for help with crystal growth and sample cutting, and
1 W. C. Nunnally and R. B. Hammond, Optoelectronic switch for pulsed power, in C. H. Lee (ed.), Picosecond Optoelectronic Devices, Orlando, FL, 1984, pp. 374-398. 2 F.V. Vald, Rev. Phys. Appl., 12 (1977) 2?7.
3(18
J. Lajzerowicz et al.
/
3 C. Canali, M. Martini, G. Ottaviani and K. R. Zanio, Phys. Rev. B, 4(1971)422. 4 F. J. Zutavern, G. Loubriel, M. W. O'Malley, L. P. Shanwald, W. D. Helgeson, D. L. McLaughlin and B. B. McKenzie, IEEE Trans. Electron. Dev., 37(1990) 2472.
Power switching with ('dTe : ('l
5 B. Schaub, J. Gallet, A. Bruno-Jailly and Pelliciari, Rev. ['hrs., 12(1977) 147. 6 N. V. Agrinskaya and O. A. Matveev, Nucl. lnstrum. Methods A, 283 (1989) 263.
Materials Science and Engineering, B I O( 1993 ! 304-308
Power switching with CdTe:C1 J. Lajzerowicz, L. Verger, F. M a t h y a n d M . C u z i n LETI (CEA, TechnologiesAvanc~es), CENG, BP 85X, 38041 Grenoble (France)
Abstract CdTe material with its high molecular weight (240) is today widely used for radiation detection. The b a n d - ~ p energy of CdTe is 1.45 eV. When doped with chlorine, which compensates the acceptor level introduced by cadmium vacancies, CdTe is intrinsic and gives a very high resistivity (more than 108f~ cm). The contacts were made by electroless metal deposition and further annealing; they were characterized with d.c. and pulsed voltages. For the first time, this paper presents power switching exp.eriments with CdTe material. We investigated different types of crystals and contact geometries with gap sizes varying from less than 1 mm to a few millimetres. The switches were activated by a YAG laser with 10 ns F W H M pulses ( 1.06/~m) or with 160 kV X-rays with 30 ns EWHM pulses. The time constant of recovery was found to be more than 10 ns. In some cases, for high voltages, corresponding to fields higher than a few kilovolts per centimetre, large recovery times of more than 100 ns were measured. This apparently Iolager carrier lifetime, combined with the high resistivity, make CdTe an alternative material to silicon and GaAs for some swishing applications.
1. Introduction--main properties of CdTe Semiconductor photoconductive switches have become increasingly useful for pulsed-power applications where high voltages must be switched on a short time-scale. Photoconductive switches can switch tens of kilovolts with picosecond rise times [1]. Most of the research has been concentrated initially on silicon and more recently on GaAs as the switch materials. Different materials, such as diamond and ZnSe, have also been evaluated for this application. To enlarge the group of potential semiconductor for the application, we decided to evaluate CdTe. CdTe has been recognized for several years as being the most promising detector material for a room temperature operating portable X-ray and gamma-ray detection system. It is also used as a substrate for epitaxial growth of H g I _xCdxTe to produce IR detectors. Semi-insulating CdTe is usually obtained by impurity compensation with chlorine or indium. Different techniques, such as the Bridgman technique or
travelling heater method (THM), are used to grow the crystals. Of course, the availability and purity of the crystals grown today cannot be c o m p ~ e d with those of silicon and GaAs. Table 1 gives the main characteristics of CdTe compared with silicon, GaAs and diamond. Different properties of CdTe must be pointed out as follows. CdTe has a direct b0JlO gap (Gunn effect is reported [2]). The mobility curve of CdT~ presents a maximum in velocity for a feld of the order of !5 kV ~m -~ [3]. Compared with GaAs and !nP, and following some of the given interpretations of the lock--on phenomenon [4], it should present the lock-on p r o p ~ y above 15 k V c m - L (The lock-on property is a non-linear effect: when the field applied is sufficiently high, instead of following the excitation pulse shape, the switch turns on and stays on until the energy is completely discharged from the generator.) CdTe recovery time is in the range right between the recovery times of silicon and GaAs.
TABLE 1. Comparison of main parameters for CdTe and the three main semiconductor materilals
Band gap (eV) Resistivity (~ cm) Mobility (cm 2 V- ~s- 1) Breakdown field (kV cm- 1) Recovery time (/~s)
0921-5107/93/$6.00
Si
AsGa
Diamond
CdTe
1.12 (indirect) 104 1000 105 10-100
1.43 (direct) 10 s 5000 105 0.001-0.01
5.47 (indirect) 10 l 1000 107 < 0.001
1.47 (direct) 109 1000 105 0.01-1
© 1993 - Elsevier Sequoia. All rights reserved
J. Lajzerowicz et al.
/
It seems very interesting to evaluate this material for power switching applications. We will briefly report how we developed ohmic contacts for our CdTe material. We will present the hold-off characteristics for impulsed voltages and, finally, we will show some results obtained when switching with laser or X-ray pulses.
2. Device preparation and characterization The monocrystalline and large volume CdTe crystals have been grown by the vertical Bridgman method by normal freezing of a tellurium-rich solution at a temperature lower than the melting point of stoichiometric CdTe [5]. Semi-insulating CdTe is obtained by impurity compensation: in the accepted scheme, a shallow donor, i.e. chlorine, is introduced into the material to compensate accepter native defects which are generally thought to be cadmium vacancies. By the selfcompensation phenomena, the balance between the chlorine donor and accepter concentration leads to Fermi level pinning within the gap that makes the material semi-insulating and thus greatly reduces photoconductor dark current [6]. Hall effect measurements exhibit a p-type conduction for the high resistivity chlorine-doped crystals. At room temperature, crystals have a resistivity p = 109 if2 cm, a hole concentration of 108 cm- 3 and a Hall mobility /~H = 50 cm 2 V-l s-1. (The Hall mobility is not the electron mobility and is much closer to the hole mobility.) We used two types of samples: one was 10 mm × 15 mm × 1 mm with the two contacts on the same large side separated by a gap of 3 mm and the other 10 mm × 10 mm x t mm (t is from 600/~m to 2 mm) with circular contacts on the opposite large sides. As with most highly intrinsic semiconductors, it is difficult to produce ohmic contacts on CdTe:C1. Our philosophy of experimentation was to carry out reliable, stable and reproducible ohmic contacts in order to obtain the lowest current for the highest electrical field applied when doing I ( V ) characteristics: 1 kV. This was obtained with electroless deposited platinum on mechanically lapped surfaces. (The surfaces are mechanically lapped to reduce the leakage current.) The solution was prepared by dissolving H2PtCI 6 in deionized water. The approximate thickness is 100 nm. Figure 1 shows the I(V) characteristics obtained for a 10 mm x 10 mm x 1 mm sample (contacts on opposite sides) before subsequent annealing. For applied voltages in the range from - 1 kV to 1 kV, the characteristic is linear and symmetric, i.e. ohmic behaviour. The sample impedance is found to be between 1 and
Power switching with CdTe.'CI
305
2 Gg2, which corresponds to a resistivity of a few gegaohm centimetres. Unfortunately, under d.c. voltages corresponding to 15 kV cm- ~most of our samples fractured. This corresponds to a rapid increase in the dark current at constant voltage. This is probably due to an increasing space charge limited conduction because of trapping on deep levels. Two different experiments were performed to test the samples under pulsed voltages. (1) Polarization by a 50 ns, 0-10 kV pulse from a line discharge (sample placed in a SF6 pressure chamber). For the 10 mm x 10 mm × 1 mm sample (contacts on opposite faces), appreciable conduction appears at 3.5 kV while at 10 kV the sample presents a 100 Q resistance. This corresponds to 20 mJ dissipated in the sample during 50 ns. For the 10 m m × 15 m m x 1 mm sample (contacts on the same side), conduction appears at 5.5 kV and the apparent resistance of the sample at 10 kV is 150 f2. After this test the I(V) characteristic of the 10 mm × 10 mm × 1 mm sample Was degraded. (2) Polarization by the 0.5/~F capacity discharge in 25 Q with 0-3 kV. The samples tested are those with contacts on opposite faces. With two contacts of the same dimensions, breakdown appears at about 22 kV
|
1,5
Y
I 0,5
--" C,)
13
-I -I ,S I~I -1000
,
.1500
11
0
I 5400
Before annealing After annealing
1000
VOLTAGE (V)
Fig. 1.1(V) characteristicsof a 10 mm × 10 mm x 1 mm sample.
Fig. 2. Sample top viewwith two breakdown paths.
306
J. Lajzerowicz et al.
/
l'ower switching with ('dl'e : (7
cm-J (1.5 kV for the 680/~m gap and 1.9 kV for the 850 # m gap). The paths are situated right along the contacts as it is shown on Fig. 2. They are holes in which the material melted. The sample shown presents two paths because, after several breakdowns in one path, the metal proceeds to sublimate and there is no more metallic continuity to the first path. The apparent breakdown resistance observed at the first breakdown at 1.5 kV (gap 680 ¢tm) is about 80 f2, which corresponds to 14 J dissipated in 10 kts, i.e. enough to melt CdTe. The paths go exactly from one contact frontier to the other contact frontier on the opposite side (see Fig. 3). The simplest solution to overcome such behaviour was to use a small contact on one face and a big contact on the opposite face (typically diameters of 2 and 7 mm on the 1 mm x 1 mm faces). Indeed, in this case, the breakdown field doubled to reach 40 kV cm- ~ (2.5 kV for the 620/~m gap and 3 kV for the 750/~m gap). The paths in this case go from the small contact frontier straight to the middle of the other contact (see Fig. 3).
dose at 1 m is 0.7 mrad. Figure 6 shows a diagram of the experimental set-up for X-ray switching under 0-3 kV pulsed voltages. The voltage pulse triggers the 300 kV high voltage X-ray tube supply; the delay is less than 1 ,us with a jitter of less than 200 ns. The time constant of the V0 pulsed voltage is 12.5/~s; therefore, during the X-ray pulse, the voltage across the switch is about 95% of V0. We only performed tests with samples with contacts on opposite faces. Switched signals give recombination times of the order of 20 ns for irradiation perpendicular to the contact and, in some conditions of irradiation from the
v
> t 206v~ I1
~
-'~-~-.
3. Results with laser or X-ray activation 200 ns/div.
We used a YAG YG 480 laser from Quantel. The energy delivered at 1.06/~m during the 12 ns pulse is 320 J. We used only d.c. polarization for laser switching tests. Different behaviours were observed in the recombination time which seems to depend on the geometry of the switch and on the field. Figure 4 gives an example of recorded switch signals at different fields. For 2.5 kVcm -~ and above, the apparent recombination times reach nearly 1/~s. Figure 5 gives the variation of the switch resistance vs. the beam attenuation at different voltages. The limit resistance seems to be lower at low fields. When the attenuation is high enough, the variation of the resistance is quasi-linear with the attenuation, with the same linearity whatever the field is. Assuming a mobility of 1000 cm 2 g -l s -1, we can calculate the equivalent energy deposited in the switch for each attenuation. We can then determine roughly the absorption of the beam to find between 0.5% and 1%. The X-ray generator is an HP 43733A flash generator. The maximum X-ray energy is 300 keV, the mean energy is 140 keV, the pulse width is 30 ns and the Same size e l e c t r ~
15"30V~
t 400 ns/div. Fig. 4. Signals from laser switching at the same attenuation but at different voltages (contacts on the same face (2 rmm gap) and laser beam on the opposite face) (recorded on Tektronix TDS 500).
10000
~-~1000 i
1 O0
#
10
Different size
1
Brea~uow. m ;~ ~ v/~=.
Breakdown at 40 K v/~m
Fig. 3. Breakdown paths and values for different contact geometries.
f, 0,05 kV/cm 0,5 kV/cm 2,5 kV/cm ........
10
t00 Beam attonuatlon
8
5 ky/cm
1000
Fig. 5. Switch resistance vs. laser beam attenuation (contacts on the same face and laser beam on the opposite face).
J. Lajzerowicz et al.
/
307
Power switching with CdTe : CI
HP 300 kV X-ray flash generator
I
10 I ~
I.tt.n..tor I ..h.°,o., I M I
"'"
I I
switch
o-3kv
/
I
.w,°,
.ttenuator
T ""T
,'°
L,';.n.°°°.,., I : TektronixI'
LI,o.o I
OSAOO'
Fig. 6. Experimental setting for switching tests with X-rays.
sides of the sample, the recombination times reach 400 ns. O n Fig. 7 is plotted the evolution of the switch resistance with the inverse of the irradiation dose at different fields. T h e irradiation dose is proportional to l i d 2 with d the distance to the source. Computation gives a carrier generation rate of 7 x 1011 c m - 3 s 1 at 1 m and, therefore, a dose of 1 corresponds to 3 × 1013 cm-3 s-1. F r o m this value and from the slope of the linear variation of the resistance, we can estimate the mobility, and we find 1150 cm 2 V-1 s-1 which is in agreement with our theory. We were able to switch at a field of 38 kV c m - l without observing any lock-on phenomenon. -
10 s ~ 104 8 ~ 103 i ~" 102 10 1
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
,o
=
.
.
.
.
loo
looo
1/(Normalized dose Irradiation)
(a)
......
L
................ / f
8i-
_I
4. C o n c l u s i o n s
T h e aim of this work was to evaluate CdTe for power switching applications. We did not obtain answers to all the questions but what we are able to say is that CdTe can withstand 40 kV c m - l and does not present any lock-on at 38 kV cm-1. This breakdown voltage must be associated with the contact technology we used and should be improved when optimizing the technology and adapting the geometries. We were surprised not to see the lock-on and this result must be taken into account by scientists who want to interpret the lock-on phenomenon.
": .... A~plie~l ~iild "=""g kV]cm ..... Applied Field = 1,2 kV/~,
102
(b)
~ A p p l i e d field = 16 kV/cm Applied field = 35 kV/cm ........ = ........
10 100 1/(Normalizeddose Irradiation)
Fig. 7. Switch resistance vs. 1/(X-ray dose irradiation). (a) D.c. voltages on i mm x 1 mm x 0.8 mm sample. (b) Impulse voltages on 1 mmx 1 mm x 0.6 mm sample (contacts on opposite faces for both samples; contact sizes are different for the second sample).
M. C. Gentet, O. Maurice and J. Rustique for technical assistance.
References Acknowledgments
T h e authors want to thank B. Schaub and G. G a u d e for help with crystal growth and sample cutting, and
1 W. C. Nunnally and R. B. Hammond, Optoelectronic switch for pulsed power, in C. H. Lee (ed.), Picosecond Optoelectronic Devices, Orlando, FL, 1984, pp. 374-398. 2 F.V. Vald, Rev. Phys. Appl., 12 (1977) 2?7.
3(18
J. Lajzerowicz et al.
/
3 C. Canali, M. Martini, G. Ottaviani and K. R. Zanio, Phys. Rev. B, 4(1971)422. 4 F. J. Zutavern, G. Loubriel, M. W. O'Malley, L. P. Shanwald, W. D. Helgeson, D. L. McLaughlin and B. B. McKenzie, IEEE Trans. Electron. Dev., 37(1990) 2472.
Power switching with ('dTe : ('l
5 B. Schaub, J. Gallet, A. Bruno-Jailly and Pelliciari, Rev. ['hrs., 12(1977) 147. 6 N. V. Agrinskaya and O. A. Matveev, Nucl. lnstrum. Methods A, 283 (1989) 263.